A LASER DOPPLER ANALYSIS OF THE IMPACT OF FLOW BOUNDARY
CONDITIONS ON THE PERFORMANCE OF A MODEL LEAN-PREMIX
COMBUSTOR
By
P.M. Anacleto and M.V. Heitor
Instituto Superior Técnico
Technical University of Lisbon
Av. Rovisco Pais, 1049-001 Lisboa; Portugal
ABSTRACT
The increasing regulation of pollutants emissions from aero-engines has stimulated the development
and test of new technologies. In particular, the demands for low NOx emissions and high combustion
efficiency has led to designs which involve the combustion of premixed fuel and air at equivalence
ratios which are close to those of the lean flammability limit. This, in turn, has led to problems of flame
stability and pressure oscillations within the combustor and associated ducting.
This paper is aimed to improve the knowledge of physics of typical flows in prevapourising and
premixing ducts and of the subsequent lean combustion process, and presents an experimental study of
the flow in a “Lean Premix Prevaporising Combustion”, LPP, gas-turbine model combustor. These
experiments, addressed the reacting flow in a laboratory model combustor at atmospheric pressure and
was intended to provide a simple case to analyse, the interactions between the flow conditions in the
fuel preparation duct, the flame stabilisation process and the downstream combusting flow, in practical
aero-engines combustors.
The experimental results obtained showed that, is possible to achieve together low NOx emissions and
high fuel efficiencies in Lean Prevaporised Premixed combustors operating at very lean conditions,
applying to high turbulence flow levels (swirl flows). The effect of the residence time is not relatively
important in Premixed lean-flames for temperatures below 1400ºC, which can be used to optimise the
fuel efficiency of the combustor. Results obtained for different flow conditions have showed that, a
change of turbulence level did not result in a change in NO x emissions.
For premixed lean-flames close to the flammability limit (temperatures below 1300ºC) the contribution
of Thermal and Fenimore NO mechanisms formation to the total NO x formed was found to be
relatively small. The N2 O pathway can be the dominant route to NO formation under these conditions.
swirl generator
50
100
100
110
air
φ=50
atomiser
φ=108
duran
glass
flameholder
liquid fuel
quartz
glass
gaseous fuel
(6 X φ 1.5mm)
Figure 1 – Schematic drawing of the test rig
100
1. INTRODUCTION
A major consideration in the development of advanced high performance gas turbine engines for
aircraft applications is to reduce pollutant emissions while increasing fuel efficiency. This is especially
true as combustor temperatures and pressures are increased in order to improve engine efficiency, since
both of these factors result in increased oxides of nitrogen (NO x) production. Significant system design
improvements are needed to reduce NOx emissions, because of their harmful effect on the earth’s
atmosphere, particularly in terms of the destruction of ozone layer in the upper atmosphere (Johnston et
al., 1989 and Wuebbles and Kinnison 1990) and the ozone formation in the lower atmosphere (which is
known as photochemical smog).
A number of strategies have been considered for reducing NOx emission from gas turbine engines
(Bahr, 1995). Of particular interest in this paper is the Lean Prevaporized Premixed (LPP) concept.
This low NOx combustor concept is based on the fact that thermal NO formation is reduced when
operating away from stoichiometric conditions. If the fuel is completely vaporized and mixed before
the flame, the overall fuel lean conditions result in a very low NOx production. In practice, however, it
is difficult to achieve a complete fuel-air mixing. As a consequence, fuel non-complete vaporization
and spatial non-uniformity and/or temporal fluctuations in the equivalence ratio, can result in locally
near-stoichiometric conditions which produce NO in excess of what is produced under perfectly mixed
conditions (Lyons, 1982 and Fric, 1992).
The development of future generations of ultra low NOx engines requires the understanding of NOformation mechanisms at very low temperatures, where Thermal NO production is not dominant
(Michaud et al., 1992 and Correa 1992). The chemistry of NOx formation in laminar flames has been
studied for a long time, and the major reaction mechanisms are well understood (Miller and Bowman,
1989). At present, the effect of turbulence on NOx formation remains unclear, some authors reports
considerable NOx reduction in turbulent premixed flames (Fujimoto et al., 1985) and others found
higher rates of NOx formation in turbulent flames when compared with laminar flames at the same
temperature (Semerjian et al., 1977).
In premixed fuel-lean flames , the contribution of the Zeldovich and the Fenimore mechanisms to the
NO x formation becomes small when the combustion temperature is below 1500ºC (Michaud and
Westmoreland 1992, Steele et al., 1995). It is proposed that most of the NOx is formed by the nitrous
oxide (N2 O) mechanism (Michaud and Westmoreland 1992, Steele et al. 1995). However, few
experimental data are available on evaluation of the mechanisms.
This paper focuses an experimental parametric study on NOx formation mechanisms in premixed leanflames, as well, the study of flow characteristics in a LPP model combustor at atmospheric pressure.
The model combustor and experimental techniques are briefly described in the following section.
Section 4 presents and discusses sample results obtained. The main conclusions of the work are
summarised in section 5.
2. EXPERIMENTAL CONDITIONS
The experimental configuration consists of a single fuel injector positioned in the centre line of a
cylindrical premixed-prevapourising duct, with flame stabilisation provided by a conical bluff-body or
by the sudden expansion at the combustor entrance. The premixing chamber is a cylindrical duct with
an inner diameter of d=50mm and a length of 150mm. The fuel, Jet-A1, is atomised by commercial
nozzles with different spray angles. Gaseous fuel can also be injected. The combustor it self, consists of
a cylindrical stainless-steel duct with an inner diameter of D=110mm and a total length of 300mm. The
combustor modules include sections that provide the necessary optical access, to allow the use of laser
techniques to visualise and quantify the inner flow.
The experiments were conducted in a test facility at atmospheric pressure. A pre-heater, deliver nonvitiated air up to T0 =500ºC, and a maximum air mass flow rate up to 0.1kg/s. The swirl generator has a
variable geometry, which allows to shift the angle of the vanes between 0º and 60º, in order to control
the swirl level of the flow. Figure 2 shows a schematic drawing of the test rig.
The experiments performed in the model combustor, allow the representation of important features of
the premixed chamber and of the primary zone of practical gas turbine combustors. The Reynolds
number, based on the mean axial velocity at the inlet section of the combustor, is about 105 . The work
involves also the analysis of the effects of incomplete premixing and prevaporisation on flame
stabilisation, propagation and pollutant formation under overall lean burning conditions.
Velocity was measured by a single component laser-Doppler velocimeter. Temperature measurements
were obtained with fine platinum-13% rhodium-platinum thermocouples. Mean concentrations of
major species were measured using a stainless steel gas sampling probe and a conventional analysers
system. Flow visualisation included the use of different techniques.
Four different flow conditions were selected for detailed studies, as described in table1. Results
presented also include parametric studies of pollutant formations.
TABLE 1 - Flow conditions for detailed combustor analysis
mair
(Kg/s)
0.05
0.05
0.05
0.05
Φ
To
(ºC)
300
300
300
300
0.5
0.5
0.5
0.5
Blade angle
(º)
45
0
45
0
s
swirl nº
0.56
0
0.64
0
Flame
stabiliser
Cone
Cone
n. a.
n. a.
Fuel
Jet A1
Jet-A1
Jet-A1
Jet-A1
Run
1
2
3
4
n.a. (not applied)
3. RESULTS AND DISCUSSION
3.1 Parametric studies
In premixed flames NO x production is a strong function of flame temperature (mainly for temperatures
above 1500ºC), which varies with the equivalence ratio and the preheating of the combustion air. The
effect of the combustor residence time in NOx production as a function of the gases temperature, are
shown in figure-2. The effect of the residence time is negligible for gases temperatures below 1500ºC.
Therefore, for these conditions the Thermal NO mechanism formation is not dominant, as for lean
premixed flames with gases temperatures above 1500ºC. The preheating of the combusting air, allows
less NOx production for similar exhaust gases temperature, which denotes the effect of the
concentration of intermediate species in the NOx formation mechanisms at low temperatures.
60
Ma ir=0.034Kg/s; To =300 K
Ma ir=0.048Kg/s; To =300 K
50
Ma ir=0.050Kg/s; To =300 K
Ma ir=0.034Kg/s; To =500 K
NOx (ppm)
40
30
Thermal NOx
20
10
0
1300
1400
1500
1600
1700
1800
Adiabatic temperature (ºC)
Figure 2 - NOx emission at combustor exit for different air mass flow rates and combustor
exhaust gases temperatures.
To address the question of which NO x mechanisms formation becomes more important at lower
temperatures, experimental data are shown in figures 3a and 3b. Figure-3a shows NOx emissions for
different exhaust gases temperatures and different levels of hydrogen dilutions. In this experiment, the
exhaust gases temperature was kept constant and increased proportions of propane has replaced by
hydrogen. The results show a decrease of the NOx emissions with the propane concentration reduction,
for temperatures above 1300ºC. These results suggest that, the Prompt NO mechanism formation
(Fenimore, 1971), which depends of the radicals HCN, have some importance in an intermediate range
of temperatures. For temperatures below 1300ºC the Prompt NO mechanism is negligible in
comparison with other or others mechanisms. Figure-3b shows NO x emission for different exhaust
gases temperatures, with propane and methane gases fuels, keeping the combusting air mass flow rate
constant. The data shows similar results of the previous figure, for temperatures below 1300ºC there is
no differences in NO x emissions between the two fuels, while for temperatures above that value the
propane shows increased emissions in comparison of methane, mainly because of the Prompt NO
mechanism (the C/H ratio is higher for the propane). These results show that, for temperatures below
1300ºC the Prompt NO mechanism formation is relatively small to the total NOx formed, then another
NO mechanism formation must be dominant under this conditions.
8
25
Adiabatic Tem peratur e
Tadi=1300ºC
Tadi=1400ºC
20
Tadi=1500ºC
Propane
NOx (p pm)
NOx (ppm)
6
4
15
Metane
10
2
5
0
0
100
90
80
70
60
1200
1300
1400
1500
1600
Toutlet (ºC)
Propane rate (dilution with hydrogen)
a)
b)
Figure 3a - NOx emissions as function of hydrogen dilution; 3b - NO x emissions for propane and
methane fuels.
According several authors, chemical mechanism models based in the N2 O pathway has shown good
agreement with experimental results obtained for fuel-lean flames at very low temperatures (Michaud
et al., 1992; Miller and Bowman, 1989 and Correa 1992). Experimental data about the effect of N2 O
addition of the flow, keeping other parameters constant showed dramatic increasing NOx emissions for
several ranges of exhaust gases temperatures, which suggest that the N2 O actuates like a radical in the
NO process formation.
Chemical
Equilibrio
equilibrium
Químico
The CO emissions, as function of the equivalence ratio are shown in figure-4. The results show the
effect of the restriction of chemistry kinetics, which prevent to achieve chemical equilibrium conditions
for low temperatures. However, this restriction can be avoided by increasing the preheating air (for the
same gases temperatures we have less fuel to react), by increasing the combustor residence time, or to
increase the turbulence intensity at the front flame (applying to swirl flows). Reviewing figure-3 we
conclude that, to increase the preheating air and keeping the exhaust gases temperature, will decrease
the NOx emissions; for gases temperatures below 1400ºC the residence time of the combustor do not
affect the NOx production. The effect of the turbulence on NOx formation will be discussed in
subsequent results (section 3.2).
100
10
Chemical
kinetics
Cinética Química
CO (ppm)
1000
Propano
Propane
Jet
Gasolina
fuel
0
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
φ
Figure 4 – CO emissions at exit combustor for different equivalence ratios
3.2 Different flow conditions studied
Figure 5 shows flame visualization in the combustor for run’s 1 to 4, respectively. The flow direction is
from the right to left side. The flow visualization exhibits incomplete fuel vaporization in the premixing
chamber for non-swirl motions (run’s 2 and 4). In these conditions the fuel droplets impinge at the wall
duct creating a liquid film, which reach the entrance of the combustor. The high air velocities at the
combustor entrance are responsible for the break up of the remaining liquid film and the consequent
fuel droplets formation, which will burn inside the combustor. However the big droplets can’t burn
completely due to the low residence time (they are convected by the high jet velocity at the combustor
inlet) and blowout of the combustor, as shown in the flow visualization.
Figure 5 – Flame visualization for run’1 to 4, respectively
The fuel spray visualization inside the premixing duct for run’3 and 4 are shown in figure 6. The left
side image shows the complete fuel vaporization in the premixing duct for run3 while the image on the
right side shows the liquid film formation downstream of the droplets impingement at the wall duct in
run4 flow condition, which reaches the combustor entrance.
Figure 6 – Liquid fuel atomisation in the premixing duct.
The mean temperature and major species concentrations fields for run’s 1to 4 are shown in figures 7 to
10, respectively. The results show a good performance (low NO x emissions together with high fuel
efficiency) of run 1and 3, due to the homogeneous mixture achieved in the premixing duct
CO (ppm)
NOx (ppm)
Figure 7 – Mean temperature and concentrations fields for run1 flow
<1ppm at exit
UHC (ppm)
200ppm at exit
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
T (ºC)
3ppm at exit
50
40
30
20
10
500
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
NOx (ppm)
220ppm at exit
CO (ppm)
3ppm at exit
UHC (ppm)
<5ppm at exit
T (ºC)
Figure 8 – Mean temperature and concentrations fields for run2 flow
CO (ppm)
215ppm at exit
UHC (ppm)
<1ppm at exit
T (ºC)
NO (ppm)
2ppm at exit
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
3
Figure 9 – Mean temperature and concentrations fields for run3 flow
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
50
40
30
20
10
0
T (ºC)
UHC (ppm)
CO (ppm)
NOx (ppm)
Figure 10 – Mean temperature and concentrations fields for run4 flow
4. CONCLUSIONS
The experimental results obtained showed that, is possible to achieve together low NOx emissions and
high fuel efficiencies in Lean Prevaporised Premixed combustors operating at very lean conditions,
applying to high turbulence flow levels (swirl flows) in the combustor.
The effect of the residence time is not relatively important in Premixed lean-flames for temperatures
below 1400ºC, which can be used to optimise the fuel efficiency of the combustor.
A good control of the equivalence ratio of the mixture and of the air preheating is necessary to assure
lean stability close to the flammability limit, as well the use of high turbulence flows (swirl flows).
For premixed lean-flames close to the flammability limit (temperatures below 1300ºC) the contribution
of Thermal and Prompt NO mechanism formation to the total NO x formed was found to be relatively
small. The N2 O pathway can be the dominant route to NO formation under these conditions.
A change in the turbulence level, between the different flows conditions, did not result in a change in
NO x emissions.
REFERENCES
Bahr, D., “Aircraft turbine engine NOx emissions abatement”. In: “Unsteady Combustion”, eds.
Culick, Heitor and Whitelaw, Kluwer Academic Publ., NATO ASI Series, Vol. E 306, pp.243-264,
1995.
Miller, J. A. and Bowman, C. T., “Mechanism and Modeling of nitrogen Chemistry in combustion”,
Prog. Energy Combust. Sci., Vol. 15, pp. 287-338, 1989.
Coats, C. M., “Comment on Review of Flashback Reported in Prevaporizing/Premixing Combustors”.
Combustion and Flame, Vol. 37, pp. 331-333, 1980.
Correa, S. M., Combustion Science and Technology, Vol. 87, pp. 329-362, 1992.
Fenimore, C. P. 13th Sympo. (Int) Combust., pp. 373-379, The Combustion Institute, Pittsburgh, 1971.
Fric, T. F., “Effects of Fuel-Air Unmixedness on NOx Emissions”, AIAA 92-3345, 28th Joint
Propulsion Conference, July 6-8, 1992.
Fujimoto, S., Kameko, Y., and Tsuruno, S., 20th Sympo. (Int.) Combust., pp. 61-66, The Combustion
Institute, Pittsburgh, 1985.
Johnston, H. S., Kinnison, D. E. and Wuebbles, D. J., “Nitrogen Oxides from High Altitude Aircraft:
An Update of Potential Effects of Ozone”, J. Geophys. Res., Vol 94, pp. 16351-16363, 1989.
Lyons, V. J., “Fuel/Air Nonuniformity-Effect on Nitric Oxide Emissions”. AIAA Journal, Vol. 20, Nº
5, pp. 660-665, 1982.
Michaud, M. G. and Westmoreland, P. R., “Chemical Mechanism of NOx Formation for Gas Turbine
Conditions”. 24th Sympo. (Int.) Combust., pp. 879-887, The Combustion Institute, Pittsburgh, 1992.
Semerjian, H. and Vranos, A., 16th Sympo. (Int.) Combust., pp. 169-179, The Combustion Institute,
Pittsburgh, 1977.
Steele, R. C., Malte, P. C., Nicol, D. G. and Kramlich, J. C., Combustion and Flame, Vol. 100, pp. 440449, 1995.
Wuebbles, D. J. and Kinnison, D. E., “Sensitivity of Strastospheric Ozone to Present and Possible
Future Aircraft Emissions”, Proceedings of a DLR International Colloquium on Air Traffic and the
Environmental – Background, Tendencies and Potential Global Atmospheric Effects, Springer-Verlag,
1990.